4.3 Sales forecasting 

Full video class on YouTube, summary and notes on Instagram, class extracts on TikTok, text below. Have fun!

The main point of this class is to understand the pros and cons of sales forecasting.

Sales forecasting refers to quantitative & qualitative techniques of predicting the future sales. The ones that we will discuss in this class are:

• Moving averages
• Variations
• Time series analysis
• Simple linear regression (from Toolkit)
• Descriptive statistics (from Toolkit)
• Market research (from 4.4)

Calculations of moving averages and variations aren’t included in exams, but you are expected to be able to interpret sales forecasting data, that is why we will learn how to work out moving averages and variations anyway. However, it is expected of you to perform sales forecasting techniques from the Toolkit (see Simple linear regression and Descriptive statistics): especially: correlation, line of best fit, extrapolation, mean, mode, median.

In the picture below, you can see two trends: one is based on the past data (grey line), and the other one os based on 3-year moving average (blue line). The former one is really “jumpy” and the latter one is quite smooth. That is exactly why moving averages are calculated: in order to identify and “smoothen” the trend in past data. Moving averages are usually based on 3 or 4 years. By the way, it does not actually have to be a “year” it can be “month” or any other time period. We can just call them “three-part moving average” in case the time period is unclear. What matters here is that you use averages of three (or four) data sets one by one in order to identify and smoothen the trend. After identifying the trend, it is then extrapolated (see Simple linear regression — Extrapolation in Toolkit) in order to predict future sales. Extrapolation is basically the extension of the current trend for the future time period.

The picture above outlines all the steps in calculating 3-year moving averages:

1. Record all past data
2. Calculate 3-year moving averages for consecutive sets of 3 years: first, calculate average sales for years 1~3, then for years 2~4, then 3~5 and so on until the last data set (8~10)
3. Put both trends on the graph: the “jumpy” past data trend and the “smooth” 3-year moving average trend
4. Extrapolate the trend by continuing the line for the future time period

For 4-year moving average it’s the same thing but you use data sets of 4 periods instead of three. Additionally, you also have to “centre” the average by calculating the averages of sets of 2 “moving averages”, but that is way beyond the IB BM syllabus, so I am not going to delve into that in my textbook. If you are interested, conduct your own research on 4-part moving averages or let me know in the comments below so that I can add them to my textbook as well.

Now, remember that your goal here is not to learn how to calculate moving averages but to be able to interpret data from them. One thing that might help you out in data interpretation is variations. Variation is the difference between the actual past data and trend (moving average). It helps to identify the periods with the highest fluctuations from the trend. This data is used to adjust marketing planning accordingly. Let’s see the table below.

As you can see from the table, the highest variations from the trend are in years 6 and 7, which indicates to the managers that they should pay particular attention to these periods. Managers might want to find out the reasons for these fluctuations so that sales forecasting is more accurate.

One more thing that managers can do with all these graphs, data sets and variations is time series analysis. Time series is simply a sequence of data (in our case, sales) recorded at regular intervals. And time series analysis refers to identifying the fluctuations and patterns in the past data and the reasons for them. Again, this is done in order to improve sales forecasting. Simply speaking, by analysing past data, managers are trying to find patterns and reasons for these patterns that will help them to predict future sales more accurately. Very often, patterns occur based on the following fluctuations: seasonal, cyclical and random.

Seasonal fluctuations are based on (surprise-surprise!) the seasons of the year. For example, ice cream consumption and its sales is higher in the summer and lower in winter. Or sales of warm coats are higher in winter.

Cyclical fluctuations are based on the stages of the business cycle — a series of economic expansion and contraction (see Figure 3 below). For example, people usually spend less money during economic recession and more money during economic growth.

And lastly, random fluctuations are unpredictable and not based on anything and do not fall under regular seasonal or economic trends. For example, if people start buying eggnog (Christmas beverage) in August, that would be a random fluctuation.

In addition to the sales forecasting techniques that we discussed above, make sure you also learn Simple linear regression and Descriptive statistics in the Toolkit: pay particular attention to extrapolation, correlation, line of best fit, mean, mode and median — they are all used for sales forecasting as well. Besides, market research is another thing that is used for sales forecasting and we will learn it later in Unit 4. Make a note of all these techniques and explore them in the corresponding sections of the textbook.

Now you are finally familiar with what sales forecasting is and you are ready to achieve the objective for this class: "discuss the benefits and limitations of sales forecasting”.

On the one hand, sales forecasting assists marketing planning: with an accurate sales forecast, managers are able to adjust the marketing mix (4Ps or 7Ps — later in Unit 4) according to the projected demand. From the financial perspective, sales forecasting helps to maintain liquidity: businesses need more cash when demand is high in order to be able to produce more goods in response to demand fluctuations. Additionally, sales forecasting helps to manage stocks: there is no need to keep excessive stocks of raw materials and components when the demand is low and, on the contrary, there is a need to have much in stock when demand is high. And lastly, sales forecasting is actually a type of planning, and as we learnt before, any type of planning reduces risk and makes businesses more prepared to any potential change.

On the other hand, sales forecasting is a technique that works under the assumption that the future is based on the past, which might not actually be true all the time… Besides, even though it is data-based, it still is just a prediction that may never come true due to unforeseen changes. Additionally, sales forecasting requires thorough work of managers and data analysts which is time-consuming and costly. And lastly, it is largely done by human beings, which means that sales forecasting is subject to bias and human factor mistakes: managers are naturally inclined to plan in their favour, so that whatever they do in the future is presented “in the good light”. For example, managers might purposefully predict low sales in order to get a bonus for exceeding the organisation’s expectations.

Make sure you can define all of these:

1. Sales forecasting
2. Moving averages
3. Three-part moving average
4. Four-part moving average
5. Extrapolation
6. Variation
7. Time series
8. Time series analysis
9. Seasonal fluctuations
10. Cyclical fluctuations
11. Business cycle
12. Random fluctuations